The magic of AI in fashion, EU’s Stalled AI Act, and hooray for Klark.ai!
💡 muffin.ai: the media to understand the challenges of AI and leverage them in our society and jobs — by a collective of French engineers & entrepreneurs.
Hi folks,
Another great week at muffin.ai!
We are continuing the new format and your voice matters a great deal to us. Share your thoughts, questions, and suggestions with us 🙏.
If anyone shared this newsletter and you’re willing to subscribe, it’s here:
On today’s agenda
🔎 AI & Fashion: How GenAI is Revolutionizing Marika d’Auteuil’s job
🍸 Snack-sized News: EU’s Stalled AI Act, Biased AI chatbots in healthcare, AI Emergence in Legal Sphere
🥗 Today's Special: Dall-E 3 integrated to ChatGPT
🇫🇷 AI Made in France: Klark.ai, the Future of Customer Service
🔎 AI & Fashion - How GenAI is Revolutionizing Marika d’Auteuil’s job
Next Tuesday, we'll be sharing our conversation with Marika D’Auteuil, a Canadian makeup artist from B. Agency. Renowned for her daring and imaginative style, Marika delves into the relationship between artificial intelligence and artistic creation.
👉 Marika's GenAI Revelation - A single snapshot sparked an exploration where Marika uncovered the magic of GenAI tools, revealing a new realm of creativity and innovation that shifted her perspective on art.
🎨 The Role of GenAI Tools in the Creative Process - For Marika, inspiration is everywhere, and GenAI tools have become invaluable allies, transforming her abstract thoughts into visual artworks, and sometimes, into unexpected creations.
⚙️ The Challenge and Nuance of Using AI Tools - Despite their accessibility, mastering GenAI tools demands significant effort to produce distinct and unique works. A process that's different, yet just as demanding as traditional photography production.
🧪 Fashion Brands' Pursuit of Innovation - For Marika, AI-generated images are a complement, not a replacement. She doesn't see professionals being overshadowed - human creations possess an authenticity, a soul, that technology can't fully replicate.
👉 For the full portrait, see you next Tuesday!
🍸 Bite-sized News
👉 Top news of the week with our take on the matter, handpicked by us.
🇪🇺 EU's Stalled AI Act - The anticipated EU AI Act faces an impasse as lawmakers struggle to agree on the regulation of foundation models. Despite prior optimism, the Act, aiming to establish a regulatory framework for generative AI, now seems unlikely to pass before the year's end. The deadlock continues, with Spain, leading the EU, advocating for a tiered regulatory system based on user numbers and regular vulnerability assessments.
💡 The proposed Act suggests rigorous processes for foundation model developers, including risk assessment, continuous testing, bias examination in training data, and pre-release documentation, indicating a high bar for compliance.
💡 The stagnation reveals a stark contrast with China, which swiftly enacted its generative AI regulations in August, and sets a precedent for international AI legislation. The delay raises concerns, especially among smaller entities and open-source communities, who urge a consideration for varying capabilities in compliance, advocating for a distinction between commercial and non-commercial developers.
🏥 AI chatbots in healthcare: a mixed blessing - A Stanford study reveals that AI chatbots, aimed to streamline healthcare, are perpetuating racial biases. The study, led by Dr. Roxana Daneshjou, found chatbots like ChatGPT and Google's Bard disseminating false racial information, especially concerning Black patients, which could potentially worsen health disparities.
💡 Despite the intent to enhance healthcare delivery, misuse or biases in AI can lead to misinformation and perpetuate existing societal issues.
💡 Companies like OpenAI and Google are actively working to reduce biases in their AI models, emphasizing that chatbots are not a substitute for professional medical advice.
🏛️ AI's Emerging Role in the Legal Sphere - As the legal sector grapples with the potential of AI, the technology presents both threats and opportunities. On one hand, AI could drastically cut jobs, as per a report from the UK's Law Society. On the other, it can significantly aid in case research and preparation, although accuracy concerns persist.
💡Ben Allgrove, Chief Innovation Officer at Baker McKenzie, emphasizes that AI's role in legal settings will likely expand through AI-powered versions of existing legal software like LexisNexis and Microsoft's 365 Solution for Legal. However, the high costs of premium AI tools, like Microsoft's Copilot, present a barrier.
💡Alex Monaco, an employment lawyer, sees AI democratizing the legal profession by aiding individuals in understanding and asserting their legal rights, especially as corporations increasingly leverage AI for various decisions. Yet, the sector is treading cautiously due to instances like the Steven Schwartz case where reliance on AI led to fabricated legal precedents, highlighting the importance of validating AI system performance.
🤖 AI Firms Liable for Harm, Experts Warn - Amidst the rapid evolution of artificial intelligence, a collective of seasoned experts including AI pioneers Geoffrey Hinton and Yoshua Bengio, underscore the urgent need for robust safety measures. The clarion call comes ahead of a crucial AI safety summit at Bletchley Park, rallying international stakeholders to mull over the emerging threats and accountability in the AI realm.
💡 A cadre of 23 experts have dubbed the unrestrained advancement of AI systems sans proper safety checks as "utterly reckless," urging for a regulatory fabric that holds AI firms accountable for any ensuing harm.
💡 The document co-authored by illustrious figures such as Stuart Russell, and Nobel laureate in economics, Daniel Kahneman, recommends earmarking a portion of AI R&D funds for ethical and safe AI development, granting auditors access to AI labs, and instituting a licensing system for cutting-edge models, amongst other measures to thwart the inadvertent descent into a dystopian AI-dominated reality.
🍎 Apple's $1 Billion AI Endeavor - Amidst a global AI boom, Apple is ramping up its investment in generative AI to the tune of $1 billion annually. The tech titan is hustling to catch up with rivals like Google, Microsoft, and Amazon who have already made significant strides in AI innovations.
💡 Apple's hefty investment aims at advancing its AI capabilities, particularly focusing on enhancing Siri, Messages, and Apple Music. The company plans to utilize AI for curating auto-generated playlists, akin to Spotify's feature, and is exploring AI integration in Xcode to aid app developers.
💡 While Apple has crafted its own large language model framework named Ajax, and a rumored chatbot dubbed Apple GPT, these technologies have not yet been incorporated into its product lineup. The initiative is spearheaded by key executives John Giannandrea, SVP of Machine Learning and AI Strategy; Craig Federighi, SVP of Software Engineering; and Eddy Cue, Head of Services.
🥗 Today's Special: Dall-E 3 integrated to ChatGPT
👉 Every week, we introduce you to practical tools and use cases that you can try out for yourself.
This week, we're spotlighting DALL-E 3, the latest version of the image generation tool, now integrated with ChatGPT.
Direct integration into ChatGPT greatly enhances the user experience, allowing for simultaneous generation of text and images from a single interface.
💡 One of DALL-E 3's standout features is its ability to comprehend detailed requests and produce images with incredible precision.
💡 Notable advancements have also been made in generating text on images, though it might take a few attempts to get the desired text.
💡 A practical tip we're sharing if you want to create images for your website:
You can try prompts like, “A professional man and woman chatting at a table with a presentation board behind them, flat simple vector illustrations style, vibrant colors, white background.”
✅ Name: DALL-E 3 via ChatGPT
📁 Description: High-precision image generation tool, directly integrated into ChatGPT
💶 Price: Available for ChatGPT Plus subscribers ($20/month)
🇫🇷 AI Made in France: Klark.ai, the Future of Customer Service
Today, we're introducing you to Klark.ai, a product launched in May 2023 that aims to revolutionize interactions between businesses and their customers through generative AI.
🛠 The Product
Klark offers a plugin for Zendesk and SalesForce, creating a unique knowledge base for each client. This base, fueled by FAQs and the history of customer interactions, generates response suggestions every time an agent opens a ticket, thereby optimizing their productivity.
Quick Setup: Installation in just a few minutes with a simple integration into Zendesk.
LLM Compatibility: The product can utilize various language models like GPT-4, GPT-3.5, Llama2, and other Open Source models on HuggingFace, thus offering better value for money.
GDPR Compliance: Servers are based in Europe, and data is pseudonymized before processing, ensuring the protection of personal data.
📈 In terms of performance, initial results show a client ticket processing time cut in half!
💡 At muffin, we love their pricing system: Klark only charges when the assistance provided is significant, a model that perfectly aligns Klark's interests with its clients'. This also reduces friction when trying out the tool.
To discover Klark or get in touch, visit their LinkedIn page.
Enjoy your week !
— muffin.ai team
If you enjoyed the edition, consider clicking on the ❤️ button and leaving a comment so that more people can discover muffin.ai on Substack 🙏